|
Abstract:
|
To meet targeted participants recruitment requirement following a scheduled timeline is crucial in the successful completion of a clinical trial. Despite increased access to data and advances in analytic techniques in the past several years, it is still a widespread problem for the delay or failure in trials due to a low rate of recruitment. The reliable and accurate prediction in site performance, as well as the consequent trial outcome, including the rate of recruitment can positively impact on trial site categorization and investments in new trials. The challenge to reliably predict the ability of a site in subject recruitment is a complicated and multifactorial problem. In this study, by taking account of real-time monitoring study accrual progression, participants status, the completion status of clinical visits, data quality such as missing forms and values, data query, as well as protocol deviations, we use LASSO regularization to determine the subset of candidate predictions that is optimal in terms of prediction accuracy in regressing the outcome on the predicators through multiple variable logistic regression analysis.
|
